空间数据挖掘是近年来迅速发展起来的面向应用的新技术。运用1998年夏季日本静止气象卫星(GMS) 探测反演出的红外辐射亮温资料和国家卫星气象中心高分辨率有限区域分析预报系统产品中的数值格点预报资料 (HLAFS) 对青藏高原上的中尺度对流系统 (MCS) 进行了自动追踪。在此基础上,运用空间关联规则研究了移出高原的MCS与其环境场之间的关系,结果表明, 在400 hPa这一层次,等压面高度、相对湿度、涡度、散度、垂直速度这5 个物理量是影响其移出高原的主要因子;而在500 hPa 层次,移出高原的MCS与等压面高度、相对湿度、温度、垂直速度和K指数关系密切。
Recently, the evidences have indicated that the heavy rainfall in Yangtze River Basin is directly caused by Mesoscale Convective System (MCS) over the Qinghai-Xizang Plateau in China.In this project, the trajectories of MCS over the Qinghai-Xizang Plateau are automatically tracked using GMS (Geostationary Meteorological Satellite) infrared black-body temperature (Tbb) and High Resolution Limited Area Analysis and Forecasting System (HLAFS) values provided by China National Satellite Meteorological Center from June to August 1998.Among these data, spatial resolution of Tbb is 0.5° lat (0.5°long, time resolution of Tbb is one hour).And, spatial resolution of HLAFS is 1° lat ( 1° long and time resolution of HLAFS is twelve hours).On the other hand, research range of latitude is from 27° to 40°N, and longitude is from 80° to 105°E.While levels include 400hPa and 500hPa.Furthermore, the study focuses on MCS that cover at least 3 connected pixels having Tbb≤-32℃ in each Tbb image, and last for at least 3 consecutive hours.Before spatial association rule mining, MCS is firstly tracked.in this project, if an MCS moves across 105°E, then the MCS is defined as "move-out of the Plateau", otherwise, it is defined as "stay in".Results indicate that there are 749 MCSs over the Qinghai-Xizang Plateau from June to August 1998.While the total number of MCS, which move out of the plateau, is 55.And, results also show that the initiation locations of most of MCSs, which moving out of the Qinghai-Xizang, are between 100°E and 105°E.Therefore, the initiation location of each MCS is defined near 100°E.On the other hand, in the course of spatial association rule mining, the parameters, isobaric surface height, temperature, relative humidity, vorticity, divergence, vertical velocity, water vapor flux divergence, θSE, K index, area, the lowest Tbb, position and shape of MCS are included.Based on these, the relationships between the trajectories of MCSs of moving out of the Qinghai-Xizang and their environmental physical field values are analysed using spatial association rule mining technique, the results indicate that at a level of 400 hPa, the MCSs, which move out of the plateau and to east, are not related with temperature, the lowest Tbb, area and location of MCS.Furthermore, the association rules, with the confidence of 1.0, are only related with isobaric surface height, relative humidity, vorticity, divergence, vertical velocity and K index.By and large, at this level, the MCS of moving out of the plateau to east is mostly determined by isobaric surface height, relative humidity, vorticity, divergence and vertical velocity.While at a level of 500 hPa, the MCSs, which moving out of the plateau, are not related with θSE, water vapor flux divergence, location and area of MCS.Furthermore, the trajectories of MCSs are less influenced by vorticity, divergence, the lowest Tbb and shape of MCS.By and large, at this level, the trajectories of MCSs are mainly influenced by isobaric surface, relative humidity, temperature, vertical velocity and Kindex.
[1] Frawley W,Piatetsk S G,Matheus C J.Knowledge discovery in databases:An Overview.in Knowledge Discovery in Databases [M].Boston,U.S.A:The Massachusetts Institute of Technology Press,1991.
[2] Fayyad U,Weir N,Djorgovski S.Automated cataloging and analysis of ski survey image databases:the SKICAT System[A].In Proc.of the Second Int.Conf.on Information and Knowledge Management[C].New York U.S.A:Association for Computing Machinery Press,1993.527-536.
[3] Koperski K,Han J.Discovery of spatial association rules in geographic information databases[A].In Egenhofer M J,Herring J R(Eds.):Advances in Spatial Databases[C].Lecture Notes in Computer Science 951(LNCS951).Springer-Verlag Press,Berlin,Germany,1995.47-66.
[4] Abraham T,Roddick J F.Opportunities for knowledge discovery in spatio-temporal information systems[J].Australian Journal of Information System,1998,5(2):3-12.
[5] Jon K,Christos P,Prabhakar R.A Microeconomic Vies of Data Mining [J].Journal of Data mining and Knowledge discovery,1998,2:311-324.,
[6] Aggarwal C C,Philip S Y.Redefining clustering for high-dimensional applications [J].IEEE Transactions on Knowledge and Data Engineering,2002,14(2):210-225.
[7] Lee R S T,Liu J N K.An automatic satellite interpretation of tropical cyclone patterns using elastic graph dynamic link model [J].IJPRAI,1999,13(8):1251-1270.
[8] Lee R S T,Liu J N K.Atmosphere - Automatic Track Mining and Objective Satellite Pattern Hunting System Using Enhanced RBF and EGDLM [J].AAAI/IAAI,2000.603-608.
[9] Kitamoto A.Spatio-temporal Data mining for typhoon image collection[J].Journal of Intelligent Information Systems,2002,19(1):25-41.
[10] Zhou Z,Chen S,Chen Z.Mining typhoon knowledge with neural networks[A].In Proc.11th Int.Conf.on Tools with Artificial Intelligence[C].The Institute of Electrical and Electronics Engineers,1999.325-326.
[11] 师春香,江吉喜,方宗义.1998长江大水期间对流云团活动特征研究 [J].气候与环境研究,2000,5(3):279~286.
[12] 江吉喜,范梅珠.夏季青藏高原上的对流云和中尺度对流系统[J].大气科学,2002,26(2):263~270.
[13] Arnaud Y,Michel D,Joel M.Automatic Tracking and characterization of African Convective Systems on Meteosat Pictures [J].Journal of Applied Meteorology,1992,31(5):443-453.
[14] Carvalho L M V,Jones C.A satellite method to identify structural properties of mesoscale convective systems based on maximum spatial correlation tracking technique (MASCOTTE) [J].Journal of Applied Meteorology,2001,40:1683-1701.
[15] http://www.cs.sfu.ca/CC/884/Wangk/kdnugget.txt.